Malicious Traffic Detection Method for Power Monitoring Systems Based on Multi-Model Fusion Stacking Ensemble...

Area of Science:

• Computer Science
• Cybersecurity
• Network Security

Background:

• Critical infrastructures like power monitoring systems face escalating cyber threats from increasing internet-based malicious traffic.
• Traditional rule-based detection methods are insufficient against complex and diverse modern network attacks.
• Existing single machine learning models for traffic detection suffer from poor generalization, low accuracy, and instability.

Purpose of the Study:

• To propose an advanced malicious traffic detection method for power monitoring systems.
• To overcome the limitations of single machine learning models in network security.
• To enhance the generalization, stability, and accuracy of malicious traffic detection.

Main Methods:

• Implemented a multi-model fusion strategy using the stacking technique.
• Integrated multiple machine learning models to create a robust detection system.
• Evaluated the proposed method on the NSL-KDD dataset.

Main Results:

• The stacking model achieved 96.5% accuracy on the NSL-KDD test set.
• The F1 score reached 96.6%, indicating high precision and recall.
• A low false-positive rate of 1.8% was recorded, demonstrating reliability.

Conclusions:

• Multi-model fusion using stacking significantly improves malicious traffic detection in power systems.
• The proposed method offers superior generalization and stability over single-model approaches.
• This research validates the effectiveness of ensemble methods in enhancing network security for critical infrastructures.